[opendtv] Re: TV Technology: 4K for Broadcast: Is it Worth the Expense?

From: Craig Birkmaier <brewmastercraig@xxxxxxxxxx>
To: opendtv@xxxxxxxxxxxxx
Date: Sat, 01 Jul 2017 09:03:00 -0400

On Jun 30, 2017, at 8:47 PM, Manfredi, Albert E
<albert.e.manfredi@xxxxxxxxxx> wrote:

http://www.tvtechnology.com/broadcast-engineering/0029/4k-for-broadcast-is-it-worth-the-expense/281218

The article makes the usual arguments, diminishing returns, viewing distance,
and so forth.

But we are grown-ups now, and we should have a sense of déjà vu. Precisely
the same arguments were made ca. 1995-2000, which is why they might sound
original to millennials.

HD became essentially mandatory, for TV screens that are commonplace today.
The naysayers back then were thinking in terms of the typical CRT-sized
displays of the day. Remember when a 35" was huge (and grainy, washed out,
and fuzzy)? In the not too distant future, we'll probably be making another
step up in size.

Yup. For a brief period in time we saw 36" CRT displays that took two people to
lift and analog projection displays with "HD capability." These were the last
gasps of a fifty year legacy with scanning CRTs that had fundamental
limitations and compromises, many of which were carried forward in the new HDTV
standard that replaced NTSC...

Interlace
59.94
Overscan
CRT Phosphor based color
A digital compression system that further limited color detail
Dynamic range limited for CRTs and 8 bit luminance

It is worth noting that Europe did not take the bait on that misstep, choosing
instead to deliver digital SDTV, which leveraged their investment in digital
"601" production gear.

What really changed?

Was it the increase in resolution that made larger displays feasible?

Or was it a breakthrough in display technology that made the poor decisions in
the ATSC standard stand out like a sore thumb?

It is well worth noting that interlace was the first "compromise" to disappear
as flat panel displays replaced CRTs. No surprise here, since interlace died
very early in displays for computers, INCLUDING those bulky analog CRT
displays. Likewise, the limited color gamut disappeared on computers as well,
at least in terms of the color encoding limits of NTSC, another issue that the
Europeans avoided way back when they chose PAL color.

The definition of color for HDTV was flawed from day one because it retained
the limits possible with the phosphors used in CRT displays. We are only now
beginning to address this issue with extended color gamut displays, like the
one I am using now on this iPad.

Sadly we had to endure SRGB (based on CRT phosphors) for the past two decades -
Thank you Mr. Poynton...

;-)

Once we understood that interlace is a crude form of image compression that has
no valid place in a digital TV system, we started making progress. Flat panel
displays first changed the world via laptop computers, then soon after for
desktop computer displays. A decade later LCD panels delivered the real
benefits of HDTV to our homes. Ironically, the Digital SDTV formats deployed in
Europe looked almost as good as the early HDTV programs delivered via ATSC.

It turned out that 720P delivered higher quality than 1080i for most forms of
TV content, especially high action movies and sports. In addition to the
breakthrough in flat panel display technology, another breakthrough in digital
inter-format conversion made the transition to flat panel displays possible.

Affordable de-interlacing chips were required to watch both the interlaced SDTV
and interlaced HDTV formats used in ATSC on inherently progressive scan flat
panel displays. And again, Europe benefitted, as their interlaced digital SDTV
formats still looked good on progressive flat panel displays.

So how did all of this influence consumer purchasing of TVs?

This chart says volumes:

This chart is for globalTV sales, so the average size in the U.S. may be
slightly larger, but the reality is that screens larger than 60 inch diagonal
are still a relatively small percentage, while the average screen size is still
hovering around 50."

And the average viewing distance has not changed appreciably in decades. This
reflects the reality that room size and furniture placement are a far more
important factor for viewing distance than screen size - the good news is that
as screen sizes get larger at the average viewing distance of 7' - 9', the
resolution requirements increase. We are now approaching the point where 1080P
resolution is optimal for most large screen displays.

But something interesting is happening with display resolution that we did not
fully understand in the early '90s when the HDTV standard was being developed.
Oversampling is beginning to play a major role, both with respect to image
acquisition AND image display.

Again, this phenomenon became obvious with computer displays, ESPECIALLY the
small displays we carry around in our pockets and purses...

The original WYSIWYG "standard" of 72 DPI is now ancient history. That number
was largely driven by the limitations of CRT displays. Now we are accustomed to
displays with more than 200 DPI, sometimes approaching 300 DPI. Why?

It's the same story as the short lived skirmish between dot matrix printers and
laser printers in the '80s. 300 DPI provided the resolution needed to print
text and graphics without obvious aliasing on edges. It is worth noting that
the increase in DPI has been accompanied by increases in screen brightness to
make this extra detail usable. This in turn has set the stage for High Dynamic
Range (HDR) displays.

But TV displays have always been different than computer displays in one
important respect - supporting alias free motion. This requires filtering to
deal with motion, which essentially cuts delivered resolution for video by
half.

The early "shorthand" for this is that interlaced 480i NTSC delivered no more
than 320 lines of vertical detail, only 240 lines to prevent worst case
aliasing. It is the reason than resolutions between 240P and 360P now dominate
video streaming to mobile devices. A computer display with 720 lines is in
essence a 360 line video display when motion/anti- alias filtering is applied.

So we have come to understand that oversampling is very important at both ends
of motion imaging systems. We use oversampled CMOS image sensors to capture
images - CMOS sensors make up for sample accuracy via oversampling and
averaging out the entropy.

And we use display oversampling to deliver both unfiltered content from the Web
and Nyquist limited content for motion video. This is where 4K comes into play.

We do not need it end-to-end in a digital television standard. But we get
better delivered image quality if we oversample during acquisition, then again
at the display.

In-between the most important factor is related to image "entropy." That is,
the accuracy of the samples that we are compressing for transport. Oversampling
during acquisition and video production, then downsampling prior to digital
video encoding delivers high quality samples, which in turn can be up-sampled
at the display without the type of image artifacts that we have been living
with in the era of bit starved MPEG-2 - i.e. over-compression producing
quantization noise and blocking artifacts.

And there is another major benefit with oversampling displays - they can ALSO
deliver unfiltered content from the web that is usable at the typical 7-9 foot
viewing distance for a TV in a home.

What all of this says is that Dr. Siebert is absolutely correct. There is no
need for 4K as an emission format moving forward.

Bert says that we are going to "grow into 4K," as consumer dislplay size
increases. From the perspective of delivering TV programming, this is highly
questionable. Perhaps some day when we all build large home theaters with >100"
screens...

But there is no valid reason to believe this will happen, in an era when HGTV
is running programs about "Tiny Houses." Europe has always understood this...

TV screen size is likely to max out around 60 inch (average) someday in the
future based on room size and aesthetics, not price.

What we need is the ability to deliver more information, not more resolution.
Brighter displays that can support HDR; increased color resolution to
reproduces all of the colors we can see. Higher frame rates to capture and
display improved motion detail.

And ALL of these improvements will improve image quality at ANY level of
resolution, be it 360P or 2048P.

Why does 1080P even exist? It was another compromise from the transition to
HDTV. Display technology is logically moving to line lengths and total lines
that take full advantage of memory configurations, not some arcane relationship
to analog video standards.

The comments to the article are more insightful than the article, I might
suggest. Don't TVs in old movies seem small and quaint? What would make
anyone think that time will stand still? A wall-sized display, in a home
setting, would likely be viewed at considerably less than three screen
heights distance. We probably should consider incorporating a 4K transmission
mode?

There was one well informed comment from my old friend Al Kovalick. We spent
years working to improve delivered image quality. Much of what I wrote above
comes from the discussions Al and I, and others who really understand how to
build end-to-end systems, have had over the past two decades.

I also liked the comment from Bill Seabrook - how about delivering decent
quality HDTV WITHOUT all of the compression artifacts?

Both John and Donald seem to agree. It's time to focus on delivered image
quality, not display size and emission formats.

The notion that we will have wall size displays is not logical. Displays of
this size will work in rooms that are designed for them - for the limited few
who can afford to dedicate a room to home theater. For the rest of us a display
that hangs on a wall, one that is aesthetically pleasing is going to become the
norm.

We don't see many homes with wall sized murals; we see rooms with artwork
hanging on the wall. A 60" to 72" display is the limit for most homes - and now
we can use that display for "artwork" when we don't need it to watch TV.

Regards
Craig

[opendtv] Re: TV Technology: 4K for Broadcast: Is it Worth the Expense?

Other related posts: